AI Paper Atlas

Research map / Large language models / Agents, code and math

FormalRx: Rectify and eXamine Semantic Failures in Autoformalization

Haocheng Wang, Baiyu Huang, Yingjia Wan, Xiao Zhu, Xiaoyang Liu, Yinya Huang, Zhijiang Guo

ICML 2026 · Large language models › Agents, code and math · approach: Inductive and abstract reasoning evaluation · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use taxonomy-guided error synthesis instead of rule-based perturbation (ablation)FormalRx-8B · FormalRx-Test verdictmacro F1: 0.100 -> 0.881Improved
use progressive training instead of joint training (ablation)FormalRx-8B · FormalRx-Test correctionaccuracy: 0.729 -> 0.792Improved
use progressive training instead of joint training (ablation)FormalRx-8B · FormalRx-Test categorizationmacro F1: 0.709 -> 0.761Improved
use progressive training instead of joint training (ablation)FormalRx-8B · FormalRx-Test verdictmacro F1: 0.881 -> 0.899Improved
use progressive training instead of joint training (ablation)FormalRx-8B · FormalRx-Test localizationaccuracy: 0.750 -> 0.778Improved
use sequential training instead of progressive replay (ablation)FormalRx-8B · FormalRx-Test unified outputoutput formatHurt
fine-tune Qwen3-8B instead of Qwen2.5-7B (ablation)FormalRx · FormalRx-Test correctionaccuracy: 0.584 -> 0.729Improved
use Qwen2.5-7B backbone instead of Qwen3-8B (ablation)FormalRx · FormalRx-Test categorizationmacro F1: 0.709 -> 0.741Improved
use R1-Distill-Qwen-7B backbone instead of Qwen3-8B (ablation)FormalRx · FormalRx-Test correctionaccuracy: 0.729 -> 0.761Improved
fine-tune on taxonomy-annotated diagnostic data vs Qwen3-8B baseQwen3-8B · FormalRx-Test categorizationmacro F1: 0.155 -> 0.709Improved
scale FormalRx from 4B to 8BFormalRx · FormalRx-Test correctionaccuracy: 0.654 -> 0.729Improved
scale FormalRx from 4B to 8BFormalRx · FormalRx-Test categorizationmacro F1: 0.508 -> 0.709Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy